Copyright © 2017 Universitas Muhammadiyah Yogyakarta - All rights reserved Journal of Electrical Technology UMY, Vol. 1, No. 1
II. Related Work
Smart machine is a machine that can understand about the task based on the command given. The
machine must have a user interface that allows users to interact with it. By nature, humans interact
and recognize his opponent by using visual sensing and vocals. Both have different characteristics and
result.
Visual sensing system is the translation of analog signals in the form of light that reflects the shape of
an object by a machine. To bring an attractive interface, a machine equipped with a camera for
detecting biometric features of the users such as face [2] and the form of the iris [3]. Visually,
translation command requires special handling because the technology is highly dependent on the
lighting, depth of image and object detection [4]. In certain cases such as the translation of complex
commands, visual sensing system will meet its limits. The use of visual sensing as interface has a
weakness, especially on accuracy are greatly affected by environmental conditions [5].
Vocal-based sensing system allows machines to understand the variation of sound provided by the
user. Voice recognition system will process voice signals into data and translate it into appropriate
speaker. Voice recognition has many variations of use such as translating voice command [6],
controlling mobile robot [7] and industrial robots [8].
In general, the voice recognition system is divided into two processes, includes feature
extraction and pattern recognition. The purpose of feature extraction is to represent the characteristics
of the speech signal by its cepstral. Cepstral represent local spectral properties of the signal for
analysis frame. Mel-frequency cepstral coefficients MFCC and bark-frequency cepstral coefficients
BFCC become candidates for spectral analysis. BFCC relatively produce better results than MFCC
in noise handling and spectral distortion [9]. Meanwhile artificial neural network is used to
identify the cepstral patterns. ANN produces better recognition accuracy rather than existing methods
[10].
III. The Proposed Method
